Biomechanical Comparison of 2 Different Locking Plate Fixation Methods in Vancouver B1 Periprosthetic Femur Fractures

Abstract

Locking plates are commonly used to treat fractures around a well-fixed femoral component. The optimal construct should provide sufficient fixation while minimizing soft-tissue dissection. The purpose of the current study was to determine whether plate length, working length, or bone mineral density affects survival of locking plate fixation for Vancouver type B1 periprosthetic hip fractures. A transverse osteotomy was created just distal to cemented femoral prostheses in 9 pairs of cadaveric femurs. Fractures were stabilized with long (20-hole) or short (12-hole) locking plates that were secured proximally with cables and screws and distally with screws only. Specimens were then cycled 10 000 times at 2500 N of axial force and 15 Nm of torque to simulate full weightbearing. A motion capture system was used to record fracture displacement during cycling. Failure occurred in 5 long and 3 short plates, with no significant differences found in the number of cycles to failure. For the specimens that survived, there were no significant differences found between long and short plates for displacement or rotation observed at the fracture site. A shorter working length was not associated with increased failure rate. Lower bone mineral density was significantly associated with failure (P = .02). We concluded that long locked plates do not appear to offer a biomechanical advantage over short locking plates in terms of fixation survival, and that bone mineral density was a better predictor of failure than was the fixation construct type.

Introduction

Periprosthetic femur fracture in association with hip arthroplasty is a serious complication that is often difficult to treat. The incidence of periprosthetic femur fracture ranges from 0.1% to 6% for primary total hip arthroplasty and is approximately 6% after revision total hip arthroplasty.^1,2 Periprosthetic fracture is currently the third most common reason, after aseptic loosening and infection, for revision surgery.¹ Given that the number of primary and revision total hip arthroplasties performed is projected to substantially increase by 2030,³ the incidence of periprosthetic fractures also is expected to rise proportionally.^4,5 Approximately 80% of periprosthetic femoral fractures occur at or just distal to the femoral stem.² Vancouver type B1 fractures are classified as those fractures that occur at or just distal to a well-fixed stem⁶ and account for approximately 29% of periprosthetic femoral fractures.² Treatment options include plates and the use of allograft struts with cables, or combinations thereof, but there is no consensus regarding the best implants or techniques to use.

Locked plating is considered to provide a more biologic fixation than standard plating techniques and has been used with success in clinical studies for type B1 fractures.^7–10 In principle, a longer plate with empty screw holes at the fracture site provides a longer working length than does a short plate but requires more extensive surgical dissection.^11,12 A longer plate on the femur also extends onto the great trochanter, possibly causing hardware prominence or affecting the vastus muscles insertion. The optimal construct for a Vancouver type B1 fracture would provide sufficient length and fixation proximally and distally while minimizing soft-tissue trauma.¹³

The purpose of the current study was to determine whether plate length, working length, or bone mineral density (BMD) affects the survival of fracture fixation. We hypothesized that, in a biomechanical cadaveric model, a longer locking plate, a longer working length, and specimens with a stronger bone density would have higher plate survival for transverse Vancouver type B1 periprosthetic hip fractures.

Materials and Methods

Specimen and Construct Preparation

Nine matched pairs (mean donor age, 79.7 ± 14.6 years) of fresh-frozen, nonembalmed human femurs (1 male and 8 females) were obtained from the State Anatomy Board and scanned for BMD using dual-energy x-ray absorptiometry (total femur mean t score –3.6 ± 0.9).

The neck cut for each specimen was standardized by using a jig and referencing from the lesser trochanter. The medullary canal was prepared for prosthetic implantation using hand reamers and broaches, and each femur was implanted with a size 3 Ultima hip prosthesis (DePuy, Warsaw, Indiana) using manually packed, acrylic bone cement (Endurance DePuy, Inc, Memphis, Tennessee).

Short- and long-plate fixation was alternated between right and left femora from pair to pair. The short plate (Figure 1A), a 12-hole stainless steel bone plate (4.5-mm Curved Broad LCP plate; Synthes, Inc, Paoli, Pennsylvania), was applied to the lateral aspect of the femur. Proximal fixation was achieved with two 1.8-mm diameter stainless steel cables and 3 unicortical locking screws (5.0-mm periprosthetic locking screw; Synthes, Inc) placed above the fracture site and 4 bicortical locking screws in the distal fragment (5.0 mm in diameter; Synthes, Inc) placed in the 4 most distal holes of the plate. The cerclage cables were secured to the plate with a cerclage locking screw that locks into the plate and were subsequently tensioned to 50 N with a tensioning device. The long plate (Figure 1B), a 20-hole stainless steel bone plate (4.5-mm Curved Broad LCP plate; Synthes, Inc), was applied to the lateral aspect of the femur with proximal and distal fixation similar to that of the short plate. The long plates required contouring to follow the flare of the greater trochanter proximally. Proximal fixation involved 2 tensioned cerclage wires, 1 placed above and 1 below the lesser trochanter, with 3 adjacent unicortical screws. All proximal screws were locked apart from the most proximal screw, which was nonlocking, so that it could be directed into the densest portion of the greater trochanter. Distal fixation was achieved with 4 evenly spaced bicortical locking screws with 1 empty plate hole between each screw. The amount of fixation with screws and cerclage wires was kept similar for the 2 groups so that the direct effect of plate length on fixation survival could be determined.

Figure 1.

Schematic to show the short (A) and long (B) plate configurations. The mean working length (WL) is defined as the distance measured between the nearest proximal and distal screws at the fracture site.

After plate application, a periprosthetic fracture was simulated by creating transverse osteotomies approximately 10-mm distal to the implant tip with an oscillating saw. A 1.3-mm section of bone (corresponding to the width of the saw blade) was removed so that a gap would be maintained at the fracture site to simulate minor fracture comminution. The distance between the nearest proximal and distal screws at the fracture site was measured using a caliper and defined as the working length.¹² The long-plate specimens had almost twice the working length (114 mm; 95% confidence interval, 111-116) of the short-plate specimens (64 mm; 95% confidence interval, 61-68 mm).

Test Procedures

The distal metaphysis of each specimen was potted in an appropriately sized section of polyvinyl chloride pipe with dental acrylic (Harry J. Bosworth Co., Skokie, Illinois) so that the mechanical axis of the femur coincided with the axis of the actuator of the MTS 858 Bionix Test System (MTS, Eden Prairie, Minnesota). Proximally, a custom-designed hinged adaptor was used to connect the trunnion of the femoral prosthesis to the load cell of the MTS machine (Figure 2 ). The adapter could rotate to allow varus/valgus motion of the proximal femur during loading.

Figure 2.

Photograph of the experimental set-up. Motion capture balls have been mounted proximal and distal to the fracture.

Specimens were loaded initially for 30 seconds to 2500 N of axial force and 15 Nm of torsion. The matched pair specimens were then cyclically loaded sinusoidally up to 10 000 cycles from 0 to 2500 N of axial force and 0 to 15 Nm of torsion (external rotation) at a rate of 1 Hz and then analyzed for failure. An optical tracking system (SMART Motion Capture System, eMotion, Inc, Padua, Italy), accurate to 0.1-mm displacement and 0.2° rotation, was used to measure motion at the fracture site. Single 3.5-mm cortical screws were inserted into the proximal and distal femur, which facilitated the mounting of tracking balls used by the motion capture system (Figure 2). Motion capture data were recorded for periods of 10 cycles at 1, 100, 500, 1000, 5000, and 10 000 cycles, unless catastrophic failure occurred before completion of the test, and the number of cycles completed was recorded. Failure was defined as obvious loss of fixation or, for those constructs that did not have obvious loss of fixation, relative rotations greater than 20°.

Statistical Analysis

Cox regression was performed to check for the effect of plate fixation type and BMD on fixation survival using Stata 10 (StataCorp LP, College Station, Texas). Specimen survival is presented as Kaplan-Meier survival curves. Effects were considered significant at P < .05.

Results

Of the 9 pairs (18 specimens), 3 short plates and 5 long plates failed (Figure 3 ). All 3 short plates failed after fracture through the distal screw holes or just distal to the plate. The 5 long plates that failed had 2 observed modes of failure: obvious loosening of fixation at the proximal femur, with screw pullout (4 plates), and plate bending at fracture site (1 plate). For the 10 specimens that survived (6 short and 4 long plates), the mean rotation at the fracture site after 10 000 cycles was 2.3° ± 1.39° for the 6 short plates and 3.3° ± 1.04° for the 4 long plates. The 1.0° greater mean rotation of the long plates was not statistically significant (P = .725). The type of plate and the working length did not significantly affect the failure rate. Bone mineral density was significantly associated with survival (P = .02; Figure 3).

Figure 3.

Kaplan Meier survival analysis for short and long plate fixations.

Discussion

Results from the current study do not support our original hypothesis that long locking plate fixations would survive longer than short locking plate fixation for transverse Vancouver type B1 periprosthetic hip fractures. Current recommendations regarding plate length include extending the plate proximally to provide overlap of the prosthesis by at least 6 screws⁹ or at least twice the outer cortical diameter of the femur.¹⁴ Some data regarding the optimal length of cortical allograft struts do exist, although the results are contradictory. Wilson et al¹⁵ conducted a small biomechanical study comparing cortical onlay allograft struts and plates for type B1 fractures. They found no difference in translation between a construct fixed with 2 20-cm strut grafts (anterior and lateral) held with cables and a similar construct using 12-cm struts. Peters et al,¹⁶ however, showed that a single 20-cm allograft strut was structurally superior to a 16-cm strut in the medial-lateral plane, but not in the anteroposterior plane, on simulated single-leg stance. Longer plates are thought to reduce plate loading, thereby avoiding fatigue failure as a result of cyclic loading.¹¹ Implants with low stiffness minimize the peak stresses at the bone-implant interface and therefore are considered more suitable for fixation in osteoporotic bone.¹⁷ According to Gautier and Sommer,¹¹ when determining the ideal length of locking plates, less than half of the plate holes should be occupied by screws. Using this method, in our study, the short (12-hole) and long (20-hole) plates had fixation-to-plate hole ratios of 0.75 and 0.45, respectively. Therefore, we did not observe the theoretical advantage of the long plate’s low screw-to-plate hole ratio.

The use of a short plate construct has several advantages for elderly, frail patients, such as less periosteal stripping proximal and distal to the fracture site and a decreased need for plate contouring around the greater trochanter. Unfortunately, the optimal stiffness to achieve biologic healing of periprosthetic femur fractures is not known. However, the implications of our study are that longer plates extending up to the greater trochanter are unnecessary for Vancouver type B1 periprosthetic fractures.

In the current biomechanical study, the working length was significantly longer in the long plate group, theoretically reducing strain at the level of the fracture site^11,12 However, a longer working length was not significantly associated with increased survival, rejecting our hypothesis that a longer working length is advantageous. A longer working length obtained by using screws further away from the fracture site may compromise the quality of proximal fixation in periprosthetic hip fractures because unicortical screw purchase in osteoporotic bone may be insufficient to maintain fracture stability, especially in the weak metaphyseal bone around the greater trochanter.

Although studies have shown that low BMD may increase the risk of periprosthetic fracture,^2,18–20 to our knowledge, it has not been established to be a significant predictor of failure after fixation. Lindahl et al²¹ attempted to determine the risk factors for failure in 1049 patients with periprosthetic femur fractures, but although osteoporosis was not identified as an independent variable, previous hip fracture and rheumatoid arthritis (often associated with osteopenia) were factors that had no significant influence on outcome.²¹ Our study shows that lower BMD is a significant risk factor of construct failure, and, therefore, treatment of periprosthetic femur fracture should include optimization of patient bone quality.

There are several limitations to this study. Although the sample size is larger than previous biomechanical studies for type B1 periprosthetic fracture fixation,^{15,16,22–24} our results should be interpreted with caution based on the small sample size used in the current study. That we were unable to identify a clear advantage of one repair over the other may be a consequence of our limited sample size. Increasing sample size would have required an unrealistic consumption of unavailable resources. To determine the effect of plate length on construct survival, we standardized the amount of proximal and distal fixation between the long and short plates, which resulted in producing a significantly longer working length for the long plates than the short plates, with empty screw holes at the level of the fracture. The use of cadaver femurs in the current study is likely to represent the heterogeneity observed in the elderly patient population, an advantage over biomechanical studies that have used synthetic femora, which are more likely to mimic higher density bone seen in younger patients.²⁵ Any variation was partially controlled for by using paired specimens. Cadaver femurs provide no biologic healing response; therefore, cyclic loading cannot take into account any additional stability created by callus formation. In conclusion, long locked plates spanning the length of the femur do not appear to offer a biomechanical advantage over short locking plates. Lower BMD was the only significant predictor of failure.